Local area networks (LAN) are communication networks operating on a common bus or which have a common central combining point, which provide intercommunications among three or more data terminals, stations or nodes. Such a network might be used, for example, for intercommunications among computer terminals in a building or on a military installation. In general, a LAN may provide signal transmission by two-wire transmission lines (as, for example, a telephone line), or for higher data rates may use coaxial cable. Significant disadvantages of coaxial cable are weight, volume and cost, which become important factors when the number of nodes to be interconnected or their separation is large. Furthermore, coaxial cable has substantial attenuation, which becomes a problem when the distances between nodes are substantial. The attenuation of coaxial cables may require the use of repeater amplifiers, which add to the cost and complexity of the communication system. Fiber optic (FO) communication networks have received increasing attention because of their advantages by comparison with coaxial cable. Fiber optic cables have the potential for being low in cost, and are light in weight and small in volume. Furthermore, the attenuation of fiber optic cables is much lower than that of coaxial cable, thereby allowing longer runs without repeaters. A further advantage of fiber optic cables over coaxial cables lies in the bandwidth, which may be greater than that of coaxial cable, thereby allowing a single fiber optic cable to carry more information than a coaxial cable. A yet further advantage of fiber-optic cables is low susceptibility to electromagnetic radiation.
In a fiber optic communication system, the computer or other source and sink of data at any node is coupled to the fiber optic bus or cable by an optical receiver with a photodetector for receiving signals, and by an optical transmitter with an optical source for transmission. The photodetector receives light signals from the fiber optic bus and converts the signals into electric signals which can be used by the computer. The optical transmitter receives electrical signals from the source of data and converts these signals into pulses of light which are propagated along the fiber optic bus.
The fiber optic transmitter may be a solid state light emitting device. The light emitting devices commonly available at the current state of the art are the light emitting diode (LED) and the laser diode. There are structural differences between LEDs and laser diodes which are not important to the invention, but functionally the differences include the fact that a LED is a relatively low power, low data rate optical source by comparison with a laser diode. For those situations in which a LAN has a high data rate, covers long distances, or has a large number of nodes (thereby requiring dividing of the transmitted light energy into many parts), the laser diode is preferred as an optical transmitter.
A laser diode has a characteristic of light output versus drive current which includes a first portion in which the light emitted as a function of drive current increases relatively slowly from zero light to a first level of light which occurs at a knee point. The characteristic also includes a second portion in which the light emitted as a function of drive current increases relatively quickly for drive currents exceeding the drive current corresponding to the knee point. The value of the maximum drive current may be established by considerations such as heating of the diode by the drive current and the resulting level of reliability. In general, for good reliability the maximum operating drive current is much less then the maximum current which the light emitting device can withstand. The data being communicated by the local area network is ordinarily in the form of binary pulses having logic HIGH and LOW levels.
In principle, the logic HIGH level of the binary pulses turn the diode to an ON condition by a forward-- bias current which causes light output which represents the logic HIGH level of the binary pulse, and the device is deenergized (deprived of current) and therefore produces no light output in order to represent a logic LOW level. For communications at low data rates this simple arrangement may be satisfactory, although the pulses may have a delay in the initiation of the optical output relative to the current drive because of delays in the conversion process of electron injection to optical (photon) emission. This in turn affects pulse symmetry and fidelity. For high data rate communications, however, timing considerations require that the transmitted light pulses have relatively well defined symmetry and overall pulse fidelity. It is known to bias each of the light emitting devices to the knee current with a direct current bias which represents a logic LOW level, and to superimpose upon the bias current a further modulating current which represents the logic HIGH level of the data to be transmitted. Thus, a logic zero level produces a light output from the light emitting device corresponding to the light output at the knee current, and the light output representing logic high level is greater than the knee current light output.
When there are a large number of nodes or stations associated with the local area network, and each node includes a light emitting device coupled to a fiber optic (FO) bus, a system problem arises if all of the light emitting devices are biased at the knee current. Each light emitting device when biased to the knee current emits a finite amount of light. Thus, the fiber optic bus has a continuous illumination attributable to the knee current of the light emitting device at each node. This illumination is a background illumination upon which the illumination representing the data to be transmitted is superimposed. The likelihood of being able to detect the desired signal decreases as the noise floor rises as a result of the background illumination. This, in turn, may reduce the maximum length of run of fiber optic cable which can be used before cable attenuation reduces the signal to an undesirably low level.
Certain topologies of local area networks allow only one node to transmit on the fiber optic bus at any one time, i.e. they use time division multiplexing of the optical bus. Each node is controlled so that it transmits at a time when the fiber optic bus is not in use by other nodes, because use by more than one transmitting node might cause the logic LOW level transmitted by a first node to be masked by the logic HIGH level transmitted by a second node, resulting in destruction of information. This scheduling of transmission is achieved in many ways known in the art, among which are "token passing" protocols in which the node which is currently transmitting, upon completion of its normal message, may transmit a signal representing the end of transmission, thereby advising the next node allowed to transmit that transmissions may begin.
U.S. Pat. No. 4,558,465 issued Dec. 10, 1985, in the name of Siegel et al. recognizes that in a communication system using light emitting devices which are biased to a knee, the cumulative light on the system represents a noise background which adversely affects communication. According to the Siegel et al. arrangement, the knee-point bias of all optical transmitters is turned off except for the one currently transmitting. When transmission is about to begin from any node, the optical transmitter at that node is biased to its knee current, and the bias is maintained for the duration of the data transmission. The data transmission is accomplished by excursions of a modulating current above the knee current. This effectively eliminates residual light on the system, and reduces system noise to improve communications, while at the same time providing desirable pulse fidelity.
The light versus bias current characteristic of laser diodes changes as a function of temperature and also of aging of the device. These changes are manifested for the most part as a change in the magnitude of the bias current at the knee between the two regions of the characteristic. Generally, the slope of light output versus bias current remains the same in the regions above and below knee point, and only the magnitude of the current at the knee point changes. Adjustment of the bias current may be accomplished by slowly increasing the bias current from zero bias current, while monitoring the light output of the light emitting device. The bias current required to obtain a predetermined light output is the knee current. In a communications system including only a single continuously operating data transmitter, it is clear that communications must be interrupted to set the bias current in this manner.
Some prior art arrangements control the bias current during data transmission by comparing the light output of logic high and logic low levels of the data with a predetermined standard, which may be the average light output. When burst data communications are involved, data bits of the burst may not be properly controlled with such a scheme because the average light output cannot be properly established due to the short duty cycle of the burst. Also, when the data rate is very high, as for example 500 Megabits/second at the current state of the art, it is difficult to sample during the data pulse to establish the amount of light existing during a logic high or low level.
In the aformentioned Siegel et al. arrangement, a control logic arrangement excites the light emitting device with a first excitation current which increases from zero. The light output of the device is monitored until the light emitted reaches the intensity representing the knee of the operating characteristic curve. A storage logic is provided for storing the magnitude of the knee current so that is can be reproduced. The knee current is then maintained constant during further bias adjustment. A second logic control circuit further excites the diode with a further increasing current until the diode emits the desired light intensity corresponding to the maximum bias. Information relating to the maximum bias current is stored so that it can also be reproduced. This completes the bias current adjustment. Switching logic is coupled to the source of data and to the logic circuits which switches the bias between the knee current bias and the maximum current bias in response to logic LOW and logic HIGH levels of the data. During the adjustment interval, the light emitting device emits the maximum amount of light. This may interfere with the communications on the network if the optical transmitter remains coupled to the fiber optic bus. Interruption of operation is required if the optical transmitter is disconnected from the fiber optic bus.
A fiber optic transmission system is desired in which the bias current of the light emitting devices may be adjusted without interrupting operation, and which is capable of use in high data rate burst mode communications systems.